.. currentmodule:: torch
Unary operators are operators that only contain only a single input. Applying them to MaskedTensors is relatively straightforward at the moment: if the data is masked out at a given index, we apply the operator, and if not, then we'll continue to mask out the data.
The available unary operators are:
.. autosummary::
:toctree: generated
:nosignatures:
abs
absolute
acos
arccos
acosh
arccosh
angle
asin
arcsin
asinh
arcsinh
atan
arctan
atanh
arctanh
bitwise_not
ceil
clamp
clip
conj_physical
cos
cosh
deg2rad
digamma
erf
erfc
erfinv
exp
exp2
expm1
fix
floor
frac
lgamma
log
log10
log1p
log2
logit
i0
isnan
nan_to_num
neg
negative
positive
pow
rad2deg
reciprocal
round
rsqrt
sigmoid
sign
sgn
signbit
sin
sinc
sinh
sqrt
square
tan
tanh
trunc
The available inplace unary operators are all of the above except:
.. autosummary::
:toctree: generated
:nosignatures:
angle
positive
signbit
isnan